adelorenzo/stable
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Logging improvements to Hyperopt (#235)

Browse Source 
* make log texts go on new line

* remove unnecessary fields from hyperopt log messages

* shorten log text in hyperopt

* consider making zero trades a failed hyperopt eval

* only log from hyperopt when result improves

* remove unnecessary temp variables

* remove unused result data variables

* remove unused import

* fix an outdated comment
This commit is contained in:
Janne Sinivirta 2017-12-25 09:18:34 +02:00 committed by Michael Egger
parent 6768658300
commit 9959d53f5e
1 changed files with 24 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

59
freqtrade/optimize/hyperopt.py
View File

@ -8,10 +8,9 @@ from functools import reduce
from math import exp
from operator import itemgetter
from hyperopt import fmin, tpe, hp, Trials, STATUS_OK
from hyperopt import fmin, tpe, hp, Trials, STATUS_OK, STATUS_FAIL
from hyperopt.mongoexp import MongoTrials
from pandas import DataFrame
import numpy as np
from freqtrade import exchange, optimize
from freqtrade.exchange import Bittrex
@ -32,9 +31,7 @@ TARGET_TRADES = 1100
TOTAL_TRIES = None
_CURRENT_TRIES = 0
TOTAL_PROFIT_TO_BEAT = 0
AVG_PROFIT_TO_BEAT = 0
AVG_DURATION_TO_BEAT = 100
CURRENT_BEST_LOSS = 100
# this is expexted avg profit * expected trade count
# for example 3.5%, 1100 trades, EXPECTED_MAX_PROFIT = 3.85
@ -100,15 +97,15 @@ SPACE = {
def log_results(results):
"if results is better than _TO_BEAT show it"
""" log results if it is better than any previous evaluation """
global CURRENT_BEST_LOSS
current_try = results['current_tries']
total_tries = results['total_tries']
result = results['result']
profit = results['total_profit']
if profit >= TOTAL_PROFIT_TO_BEAT:
logger.info('\n{:5d}/{}: {}'.format(current_try, total_tries, result))
if results['loss'] < CURRENT_BEST_LOSS:
CURRENT_BEST_LOSS = results['loss']
logger.info('{:5d}/{}: {}'.format(
results['current_tries'],
results['total_tries'],
results['result']))
else:
print('.', end='')
sys.stdout.flush()
@ -127,37 +124,37 @@ def optimizer(params):
total_profit = results.profit_percent.sum()
trade_count = len(results.index)
if trade_count == 0:
return {
'status': STATUS_FAIL,
'loss': float('inf')
}
trade_loss = 1 - 0.35 * exp(-(trade_count - TARGET_TRADES) ** 2 / 10 ** 5.2)
profit_loss = max(0, 1 - total_profit / EXPECTED_MAX_PROFIT)
loss = trade_loss + profit_loss
_CURRENT_TRIES += 1
result_data = {
'trade_count': trade_count,
'total_profit': total_profit,
'trade_loss': trade_loss,
'profit_loss': profit_loss,
'avg_profit': results.profit_percent.mean() * 100.0,
'avg_duration': results.duration.mean() * 5,
'loss': loss,
'current_tries': _CURRENT_TRIES,
'total_tries': TOTAL_TRIES,
'result': result,
'results': results
}
log_results(result_data)
return {
'loss': trade_loss + profit_loss,
'loss': loss,
'status': STATUS_OK,
'result': result,
'total_profit': total_profit,
'avg_profit': result_data['avg_profit'],
'avg_profit': results.profit_percent.mean() * 100.0,
}
def format_results(results: DataFrame):
return ('Made {:6d} buys. Average profit {: 5.2f}%. '
'Total profit was {: 11.8f} BTC. Average duration {:5.1f} mins.').format(
return ('{:6d} trades. Avg profit {: 5.2f}%. '
'Total profit {: 11.8f} BTC. Avg duration {:5.1f} mins.').format(
len(results.index),
results.profit_percent.mean() * 100.0,
results.profit_BTC.sum(),
@ -165,10 +162,6 @@ def format_results(results: DataFrame):
)
def filter_nan(result, filter_key):
return [r for r in result if not np.isnan(r[filter_key])]
def buy_strategy_generator(params):
def populate_buy_trend(dataframe: DataFrame) -> DataFrame:
conditions = []
@ -223,7 +216,7 @@ def start(args):
# Initialize logger
logging.basicConfig(
level=args.loglevel,
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
format='\n%(message)s',
)
logger.info('Using config: %s ...', args.config)
@ -244,9 +237,5 @@ def start(args):
best = fmin(fn=optimizer, space=SPACE, algo=tpe.suggest, max_evals=TOTAL_TRIES, trials=trials)
logger.info('Best parameters:\n%s', json.dumps(best, indent=4))
filt_res = filter_nan(trials.results, 'total_profit')
filt_res = filter_nan(filt_res, 'avg_profit')
results = sorted(filt_res, key=itemgetter('loss'))
results = sorted(trials.results, key=itemgetter('loss'))
logger.info('Best Result:\n%s', results[0]['result'])